@ailib-official/ai-protocol 0.8.4

package/dist/v1/providers/hunyuan.json

@@ -0,0 +1,198 @@
+{
+  "$schema": "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json",
+  "id": "hunyuan",
+  "protocol_version": "1.5",
+  "name": "腾讯混元 (Tencent Hunyuan)",
+  "version": "v1",
+  "status": "stable",
+  "category": "ai_provider",
+  "official_url": "https://cloud.tencent.com/document/product/1729",
+  "support_contact": "https://cloud.tencent.com/online-service",
+  "endpoint": {
+    "base_url": "https://hunyuan.tencentcloudapi.com",
+    "protocol": "https",
+    "timeout_ms": 60000
+  },
+  "auth": {
+    "type": "bearer",
+    "token_env": "HUNYUAN_API_KEY"
+  },
+  "payload_format": "openai_style",
+  "api_families": [
+    "chat_completions"
+  ],
+  "default_api_family": "chat_completions",
+  "endpoints": {
+    "chat": {
+      "path": "/v1/chat/completions",
+      "method": "POST",
+      "adapter": "openai"
+    }
+  },
+  "termination": {
+    "source_field": "finish_reason",
+    "mapping": {
+      "stop": "end_turn",
+      "length": "max_tokens",
+      "tool_calls": "tool_use",
+      "sensitive": "refusal"
+    },
+    "notes": [
+      "OpenAI-compatible finish_reason.",
+      "混元大模型 (Hunyuan) is Tencent's flagship LLM."
+    ]
+  },
+  "tooling": {
+    "source_model": "openai_tool_calls",
+    "tool_use": {
+      "id_path": "id",
+      "name_path": "function.name",
+      "input_path": "function.arguments",
+      "input_format": "json_string"
+    },
+    "notes": [
+      "Supports OpenAI-compatible function calling."
+    ]
+  },
+  "retry_policy": {
+    "strategy": "exponential_backoff",
+    "max_retries": 3,
+    "min_delay_ms": 1000,
+    "max_delay_ms": 30000,
+    "jitter": "full",
+    "retry_on_http_status": [
+      429,
+      500,
+      502,
+      503
+    ]
+  },
+  "error_classification": {
+    "by_http_status": {
+      "400": "invalid_request",
+      "401": "authentication",
+      "403": "permission_denied",
+      "404": "not_found",
+      "429": "rate_limited",
+      "500": "server_error",
+      "503": "overloaded"
+    }
+  },
+  "parameter_mappings": {
+    "temperature": "temperature",
+    "max_tokens": "max_tokens",
+    "stream": "stream",
+    "top_p": "top_p",
+    "tools": "tools",
+    "tool_choice": "tool_choice"
+  },
+  "response_format": "openai_style",
+  "response_paths": {
+    "content": "choices[0].message.content",
+    "tool_calls": "choices[0].message.tool_calls",
+    "usage": "usage",
+    "finish_reason": "choices[0].finish_reason"
+  },
+  "streaming": {
+    "event_format": "data_lines",
+    "decoder": {
+      "format": "sse",
+      "delimiter": "\n\n",
+      "prefix": "data: ",
+      "done_signal": "[DONE]"
+    },
+    "content_path": "choices[0].delta.content",
+    "tool_call_path": "choices[0].delta.tool_calls",
+    "usage_path": "usage",
+    "frame_selector": "exists($.choices) || exists($.error)",
+    "event_map": [
+      {
+        "match": "exists($.choices[*].delta.content)",
+        "emit": "PartialContentDelta",
+        "fields": {
+          "content": "$.choices[*].delta.content"
+        }
+      },
+      {
+        "match": "exists($.choices[*].delta.tool_calls[*].function.name)",
+        "emit": "ToolCallStarted",
+        "fields": {
+          "tool_call_id": "$.choices[*].delta.tool_calls[*].id",
+          "tool_name": "$.choices[*].delta.tool_calls[*].function.name",
+          "index": "$.choices[*].delta.tool_calls[*].index"
+        }
+      },
+      {
+        "match": "exists($.choices[*].delta.tool_calls[*].function.arguments)",
+        "emit": "PartialToolCall",
+        "fields": {
+          "arguments": "$.choices[*].delta.tool_calls[*].function.arguments",
+          "index": "$.choices[*].delta.tool_calls[*].index"
+        }
+      },
+      {
+        "match": "exists($.usage)",
+        "emit": "Metadata",
+        "fields": {
+          "usage": "$.usage"
+        }
+      },
+      {
+        "match": "exists($.choices[*].finish_reason)",
+        "emit": "FinalCandidate",
+        "fields": {
+          "finish_reason": "$.choices[*].finish_reason"
+        }
+      }
+    ],
+    "stop_condition": "$.choices[0].finish_reason != null"
+  },
+  "features": {
+    "multi_candidate": {
+      "support_type": "native",
+      "param_name": "n"
+    },
+    "response_mapping": {
+      "tool_calls": {
+        "path": "choices[0].message.tool_calls",
+        "fields": {
+          "id": "id",
+          "name": "function.name",
+          "args": "function.arguments"
+        }
+      },
+      "error": {
+        "message_path": "error.message",
+        "code_path": "error.code"
+      }
+    }
+  },
+  "capabilities": {
+    "streaming": true,
+    "tools": true,
+    "vision": true,
+    "agentic": true,
+    "parallel_tools": true,
+    "reasoning": true
+  },
+  "availability": {
+    "required": false,
+    "regions": [
+      "cn"
+    ],
+    "check": {
+      "method": "GET",
+      "path": "/v1/models",
+      "expected_status": [
+        200,
+        401
+      ],
+      "timeout_ms": 5000
+    }
+  },
+  "experimental_features": [
+    "search_enhancement",
+    "image_understanding",
+    "role_play"
+  ]
+}

package/dist/v1/providers/jina.json

@@ -0,0 +1,202 @@
+{
+  "$schema": "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json",
+  "id": "jina",
+  "protocol_version": "1.5",
+  "name": "Jina AI",
+  "version": "v1",
+  "status": "stable",
+  "category": "ai_provider",
+  "official_url": "https://jina.ai/docs",
+  "support_contact": "https://jina.ai/contact",
+  "endpoint": {
+    "base_url": "https://api.jina.ai",
+    "protocol": "https",
+    "timeout_ms": 30000
+  },
+  "auth": {
+    "type": "bearer",
+    "token_env": "JINA_API_KEY",
+    "payload_format": "jina_api"
+  },
+  "api_families": [
+    "embeddings",
+    "rerank",
+    "reader",
+    "search"
+  ],
+  "default_api_family": "embeddings",
+  "endpoints": {
+    "embeddings": {
+      "path": "/v1/embeddings",
+      "method": "POST",
+      "adapter": "openai"
+    },
+    "rerank": {
+      "path": "/v1/rerank",
+      "method": "POST",
+      "adapter": "jina"
+    },
+    "reader": {
+      "path": "/v1/reader",
+      "method": "POST",
+      "adapter": "jina"
+    }
+  },
+  "services": {
+    "list_models": {
+      "path": "/v1/models",
+      "method": "GET",
+      "response_binding": "data"
+    }
+  },
+  "termination": {
+    "source_field": "finish_reason",
+    "mapping": {
+      "stop": "end_turn",
+      "length": "max_tokens"
+    },
+    "notes": [
+      "Jina Reader uses finish_reason for completion",
+      "Embeddings API does not use finish_reason"
+    ]
+  },
+  "tooling": {
+    "source_model": "unknown",
+    "notes": [
+      "Jina AI is primarily embeddings and search",
+      "Does not support traditional tool calling"
+    ]
+  },
+  "retry_policy": {
+    "strategy": "exponential_backoff",
+    "max_retries": 3,
+    "max_delay_ms": 30000,
+    "min_delay_ms": 1000,
+    "jitter": "full",
+    "retry_on_http_status": [
+      429,
+      500,
+      502,
+      503
+    ],
+    "notes": [
+      "Jina API is generally fast but may experience spikes",
+      "Exponential backoff recommended for rate limits"
+    ]
+  },
+  "error_classification": {
+    "by_http_status": {
+      "400": "invalid_request",
+      "401": "authentication",
+      "403": "permission_denied",
+      "404": "not_found",
+      "429": "rate_limited",
+      "500": "server_error",
+      "502": "overloaded",
+      "503": "overloaded"
+    },
+    "notes": [
+      "Jina returns clear error messages in JSON format",
+      "Rate limit info included in error responses"
+    ]
+  },
+  "parameter_mappings": {
+    "model": "model",
+    "input": "input",
+    "encoding_format": "encoding_format",
+    "dimensions": "dimensions",
+    "top_n": "top_n",
+    "query": "query",
+    "documents": "documents",
+    "return_documents": "return_documents",
+    "url": "url",
+    "question": "question"
+  },
+  "response_format": "jina_style",
+  "response_paths": {
+    "content": "output",
+    "embedding": "data.embedding",
+    "data": "data",
+    "results": "results"
+  },
+  "streaming": {
+    "event_format": "jina_stream",
+    "decoder": {
+      "format": "sse",
+      "delimiter": "\n\n",
+      "prefix": "data: ",
+      "done_signal": "[DONE]"
+    },
+    "frame_selector": "exists($.delta) || exists($.embedding) || exists($.results)",
+    "event_map": [
+      {
+        "match": "exists($.delta)",
+        "emit": "PartialContentDelta",
+        "fields": {
+          "content": "$.delta.content"
+        }
+      },
+      {
+        "match": "exists($.embedding)",
+        "emit": "Metadata",
+        "fields": {
+          "embedding": "$.embedding",
+          "index": "$.index"
+        }
+      },
+      {
+        "match": "exists($.results)",
+        "emit": "FinalCandidate",
+        "fields": {
+          "results": "$.results"
+        }
+      }
+    ],
+    "stop_condition": "exists($.finish_reason) || exists($.results)",
+    "extra_metadata_path": "$.usage"
+  },
+  "features": {
+    "multi_candidate": {
+      "support_type": "native",
+      "param_name": "top_n"
+    },
+    "response_mapping": {
+      "error": {
+        "message_path": "message",
+        "code_path": "code",
+        "type_path": "error"
+      }
+    }
+  },
+  "capabilities": {
+    "streaming": false,
+    "tools": false,
+    "vision": false,
+    "agentic": false,
+    "parallel_tools": false,
+    "reasoning": false
+  },
+  "experimental_features": [
+    "rerank_api",
+    "reader_api",
+    "multimodal_embeddings",
+    "code_embeddings"
+  ],
+  "availability": {
+    "required": false,
+    "regions": [
+      "global"
+    ],
+    "check": {
+      "method": "GET",
+      "path": "/v1/models",
+      "expected_status": [
+        200
+      ],
+      "timeout_ms": 3000
+    },
+    "notes": [
+      "Check v1/models endpoint to verify service health"
+    ]
+  }
+}

package/dist/v1/providers/lepton.json

@@ -0,0 +1,221 @@
+{
+  "$schema": "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json",
+  "id": "lepton",
+  "protocol_version": "1.5",
+  "name": "Lepton AI",
+  "version": "v1",
+  "status": "stable",
+  "category": "ai_provider",
+  "official_url": "https://www.lepton.ai/docs",
+  "support_contact": "https://www.lepton.ai/contact",
+  "endpoint": {
+    "base_url": "https://llama3-1-8b.lepton.run/api/v1",
+    "protocol": "https",
+    "timeout_ms": 30000
+  },
+  "base_url_template": "https://{model_id}.lepton.run/api/v1",
+  "auth": {
+    "type": "bearer",
+    "token_env": "LEPTON_API_TOKEN"
+  },
+  "payload_format": "openai_style",
+  "api_families": [
+    "chat_completions",
+    "completions"
+  ],
+  "default_api_family": "chat_completions",
+  "endpoints": {
+    "chat": {
+      "path": "/chat/completions",
+      "method": "POST",
+      "adapter": "openai"
+    },
+    "completions": {
+      "path": "/completions",
+      "method": "POST",
+      "adapter": "openai"
+    }
+  },
+  "services": {
+    "list_models": {
+      "path": "/models",
+      "method": "GET",
+      "response_binding": "data"
+    }
+  },
+  "termination": {
+    "source_field": "finish_reason",
+    "mapping": {
+      "stop": "end_turn",
+      "length": "max_tokens",
+      "tool_calls": "tool_use"
+    },
+    "notes": [
+      "OpenAI-compatible finish_reason."
+    ]
+  },
+  "tooling": {
+    "source_model": "openai_tool_calls",
+    "tool_use": {
+      "id_path": "id",
+      "name_path": "function.name",
+      "input_path": "function.arguments",
+      "input_format": "json_string"
+    },
+    "notes": [
+      "OpenAI-compatible tool calling for supported models."
+    ]
+  },
+  "retry_policy": {
+    "strategy": "exponential_backoff",
+    "max_retries": 3,
+    "min_delay_ms": 1000,
+    "max_delay_ms": 30000,
+    "jitter": "full",
+    "retry_on_http_status": [
+      429,
+      500,
+      502,
+      503
+    ],
+    "notes": [
+      "Standard exponential backoff for rate limits and transient errors."
+    ]
+  },
+  "error_classification": {
+    "by_http_status": {
+      "400": "invalid_request",
+      "401": "authentication",
+      "403": "permission_denied",
+      "404": "not_found",
+      "429": "rate_limited",
+      "500": "server_error",
+      "502": "server_error",
+      "503": "overloaded"
+    }
+  },
+  "parameter_mappings": {
+    "temperature": "temperature",
+    "max_tokens": "max_tokens",
+    "stream": "stream",
+    "top_p": "top_p",
+    "stop_sequences": "stop",
+    "frequency_penalty": "frequency_penalty",
+    "presence_penalty": "presence_penalty",
+    "tools": "tools",
+    "tool_choice": "tool_choice",
+    "seed": "seed",
+    "top_k": "top_k"
+  },
+  "response_format": "openai_style",
+  "response_paths": {
+    "content": "choices[0].message.content",
+    "tool_calls": "choices[0].message.tool_calls",
+    "usage": "usage",
+    "finish_reason": "choices[0].finish_reason"
+  },
+  "streaming": {
+    "event_format": "data_lines",
+    "decoder": {
+      "format": "sse",
+      "delimiter": "\n\n",
+      "prefix": "data: ",
+      "done_signal": "[DONE]"
+    },
+    "content_path": "choices[0].delta.content",
+    "tool_call_path": "choices[0].delta.tool_calls",
+    "usage_path": "usage",
+    "frame_selector": "exists($.choices) || exists($.error)",
+    "event_map": [
+      {
+        "match": "exists($.choices[*].delta.content)",
+        "emit": "PartialContentDelta",
+        "fields": {
+          "content": "$.choices[*].delta.content"
+        }
+      },
+      {
+        "match": "exists($.choices[*].delta.tool_calls[*].function.name)",
+        "emit": "ToolCallStarted",
+        "fields": {
+          "tool_call_id": "$.choices[*].delta.tool_calls[*].id",
+          "tool_name": "$.choices[*].delta.tool_calls[*].function.name",
+          "index": "$.choices[*].delta.tool_calls[*].index"
+        }
+      },
+      {
+        "match": "exists($.choices[*].delta.tool_calls[*].function.arguments)",
+        "emit": "PartialToolCall",
+        "fields": {
+          "arguments": "$.choices[*].delta.tool_calls[*].function.arguments",
+          "index": "$.choices[*].delta.tool_calls[*].index"
+        }
+      },
+      {
+        "match": "exists($.usage)",
+        "emit": "Metadata",
+        "fields": {
+          "usage": "$.usage"
+        }
+      },
+      {
+        "match": "exists($.choices[*].finish_reason)",
+        "emit": "FinalCandidate",
+        "fields": {
+          "finish_reason": "$.choices[*].finish_reason",
+          "candidate_index": "$.choices[*].index"
+        }
+      }
+    ],
+    "stop_condition": "$.choices[0].finish_reason != null"
+  },
+  "features": {
+    "multi_candidate": {
+      "support_type": "native",
+      "param_name": "n"
+    },
+    "response_mapping": {
+      "tool_calls": {
+        "path": "choices[0].message.tool_calls",
+        "fields": {
+          "id": "id",
+          "name": "function.name",
+          "args": "function.arguments"
+        }
+      },
+      "error": {
+        "message_path": "error.message",
+        "code_path": "error.code",
+        "type_path": "error.type"
+      }
+    }
+  },
+  "capabilities": {
+    "streaming": true,
+    "tools": true,
+    "vision": true,
+    "agentic": true,
+    "parallel_tools": true,
+    "reasoning": false
+  },
+  "availability": {
+    "required": false,
+    "regions": [
+      "global"
+    ],
+    "check": {
+      "method": "GET",
+      "path": "/models",
+      "expected_status": [
+        200,
+        401
+      ],
+      "timeout_ms": 5000
+    }
+  },
+  "experimental_features": [
+    "photon_deployment",
+    "auto_scaling",
+    "custom_models"
+  ]
+}